Upload
others
View
6
Download
0
Embed Size (px)
Citation preview
Realistic Physics for Dynamic Rupture Scenarios: The Example of the 1992 Landers EarthquakeStephanie Wollherr, Alice-Agnes Gabriel and Heiner Igel, Department of Earth and Environmental Sciences, LMU Munich. Contact: [email protected]
INTRODUCTION MODEL OF THE 1992 LANDERS EARTHQUAKE SOFTWARE PACKAGE SEISSOL
References: [1] Fleming et al. (1998): Harris et al. (2009): Fractures along a portion of the Emerson fault zone related to the 1992 Landers, California, earthquake: Evidence for the rotation of the Galway-Lake-Road block, Geol. Society of America. [2] National Geospatial-Intelligence Agency (2000): SRTM3 Digital Elevation Model, Version 2.1. [3] Madden & Pollard (2012): Integration of surface slip and aftershocks to constrain the 3D structure of faults involved in the M 7.3 Landers earthquake, BSSA. [4] Graves & Pitarka (2010): Broadband Ground-Motion Simulation Using a Hybrid Approach, BSSA. [5] Roten et al. (2015): Quantification of fault zone plasticity effects with spontaneous rupture simulations, BEST PSHANI. [6] Pelties, Gabriel, Ampuero (2014): Verification of an ADER-DG Method for Complex Dynamic Rupture Problems, Geosc. Model Develop. [7] The SCEC/USGS Dynamic Earthquake Rupture Code Verification Exercise, Seismol. Res. Lett. [8] Barall (2009): A grid-doubling finite-element technique for calculating dynamic three-dimensional spontaneous rupture on an earthquake fault, GJI. [9] Breuer et al. (2014): Sustained Petascale Performance of Seismic Simulations with SeisSol on SuperMUC, ISC '14. [10] Heinecke et al. (2014): Petascale High Order Dynamic Rupture Earthquake Simulations on Heterogeneous Supercomputers, SC '14. [11] Rettenberger et al. (2015): Optimizing Large Scale I/O for Petascale Seismic Simulations on Unstructured Meshes, IEEE '15. [12] Rettenberger et al. (2016): ASAGI - A Parallel Solver for Adaptive Geoinformation, ACM. [13] Milliner et al. (2016): Comparison of coseismic near-field and off-fault surface deformation patterns of the 1992 Mw 7.3 Landers and 1999 Mw 7.1 Hector Mine earthquakes: Implications for controls on the distribution of surface strain, GRL.
LARGE-SCALE SIMULATION WITH OFF-FAULT PLASTICITY (Wollherr & Gabriel, in prep.)
SeisSol is an open source software package for wave propagation and dynamic rupture (DR) simulations solving the (visco-) elastic wave equation:
coupled to a frictional boundary conditions on the fault interface.→ verified for numerous DR problems [6] using SCEC benchmarks [7] including branched faults, rate-and-state friction, off-fault plasticity...
Main characteristics
ADER-DG method unstructured tetrahedral meshes
(complex geometries, topography, rough faults) highly-optimized for the use on heterogeneous supercomputers [8]; petaflop performance on several supercomputers worldwide and
Gordon Bell prize finalist 2014 [9] optimized I/O routines: parallel customized mesh reader based on
netcdf [10], high-resolution wave field output, ASAGI for fast input of 3D velocity structures [11]
To model realistic earthquake source dynamics, many ingredients must be taken into account: On the one hand, the fault geometry and the initial stress setup exert large influences on rupture propagation. On the other hand, it is crucial to use realistic physical assumptions such as off-fault plasticity. Here we present a dynamic rupture simulation of the 1992 Landers earthquake including a complex branched fault geometry, topography and off-fault plasticity.We show how plastic yielding influences the spatio-temporal source characteristics and compare the off-fault deformation pattern of our simulation with field observations.
The 1992 Landers earthquake
34°N13' and 116°W26' at 1158 UT, June 28, 1992, Mw = 7.3
prominent example of an earthquake on a complex fault structure:rupture propagation along 5 different faults, rupture jumps across several releasing step oversbi-lateral propagation at the Homestead-Valley Fault Results of SeisSol compared to FaultMod [7]
for an elastic (e) and plastic (p) strike-slip scenario for TPV27.
CONCLUSION
plasticity changes the overall spatio-temporal rupture transfer between fault segments
- rupture jump delay- rupture jump at different location- prohibits branching
reduction of the peak slip rate up to 50%
locally higher slip
plastic energy dissipation prevents surface rupture
accumulated plastic strain shows features of observed damaged fault zone, specifically at complex structures such as step-over zones between segments
fault traces based on photometric images [1], extended 16km in depth
regional topography from digital elevation models (SRTM3, [2])
lateral homogeneous but depth-dependent initial stress,direction of principal stress: N11°E [3]
1D velocity structure [4]
Linear slip-weakening (LSW) friction law;same friction parameters for the whole fault
depth dependent plastic cohesion based on [5] (Hoek-Brown parameter for granite)
→ very hard to constrain cohesion
● 100m resolution on the fault● unstructured grid,
coarsening away from the fault ~ 6%● 22 million elements● 4*10 degrees of freedom⁹
Three-dimensional model of the Landers scenario: Fault segments embedded in a realistic geological structure including topography.
Top: Depth-dependent initial stresses, plastic cohesion and velocity structure.
Ground motion of the Landers scenario in termsof absolute particle velocity and four seismogramsin the vicinity of the Landers fault.
Comparison slip rate over time
elastic:first rupture jump to the HVF at ~7splastic: no jump at ~7s but at ~7.9s at a more distant location
elastic:jump to EF and branching from HVFplastic:jump to EF, no branching
elastic:multiple rupture fronts at EFplastic:very low slip rates at EF and CRF due to previous energy dissipation,delayed rupture initiation at CRF
General features in both simulations
● complex rupture propagation● back propagating rupture● rupture jumps
● rupture duration ~25s● total slip comparable to Landers
1300 nodes à 16 OMP threadson SuperMUC Phase 130 seconds simulated time+ fault output (1s) + wavefield/strain output (5s)+ 60 fault receiver + 380 seismometers
elastic: 2h22minplastic: 7h18min CRF – Camp Rock Fault
EF – Emerson FaultCF – Connector Fault between RF and HVF
time
JVFHVFEF
CRF
LKFelastic
7.9 s
9.3 s
11.3 s
12.1 s
20.4 s
7.1 s
slip rate m/splasticCF
r2r1
slip rate m/s
HVF – Homestead Valley FaultLKF – Landers Kickapoo FaultJVR – Johnson Valley Fault
cohesionfriction anglemean stressdeviatoric stress
elastic calculations of all unknowns
elastic response plastic response
adjust stresses→ increase in plastic strain
check yield criterion
Plasticity Implementation
Return map algorithm with Drucker-Prager yield criterion r1
r2
Distribution of final slip: more elements with higher slip in the plastic simulation
Slip rate reduction
Final Slip top: elastic bottom: plastic
www.seissol.org
Milliner et al. (2016) [13]“... FZW increases where the rupture becomes structurally complex (e.g., at branches, bends, or terminations).”
Accumulated plastic strain from our simulationis confined in a narrow zone around the fault
Fault Zone Width (FZW) measurements by Milliner et al.(2016)
Comparison with observations
Can we constrain parameters such as cohesion by comparing plastic strain in our simulation to observations?
Accumulated plastic strain:
increment of inelastic strain
rupturejump locations
OUTLOOK
VELOCITY WEAKENING MECHANISMS
Rate-and-State friction with velocity weakening
Thermal pressurization (TP)frictional heating due to rapid slip increases pore pressure → reduction of the effective normal stress
Verification and Convergence
verified by SCEC benchmark tests
Convergences tests show: - robust peak slip rate, - but high-resolution needed for accurately capture rupture time!
implementation benchmarked (TPV105)
next: incorporation in realistic 3D setups
without TP with TP
fault along-strike distance
→ sustained rupture due to thermal weakening
rupture dies out